Chapter 17: Arconic's testing of ACM panels 1997 to 2006

¶17.1 The fire testing carried out by Arconic on various forms of ACM panels between 1997 and 2006 forms an important part of the background to its application for a BBA certificate. It is also relevant to the way in which it promoted Reynobond 55 PE in the years that followed.

January 1997: Reynobond 160 PE

¶17.2 Before the early 2000s, Arconic did not produce ACM panels with polyethylene cores. It manufactured coated aluminium sheets at its factory at Merxheim, in France, which it then sent to a sister company in the United States, Arconic Architectural Products LLC, previously Reynolds Metals Company, which bonded them to a polyethylene core made in the United States to form sheets of composite material. The finished sheets were then transported back to Arconic for storage at Merxheim.⁹⁹ Schmidt {Day91/35:7-13}. We understand that the product, which was sold by Arconic from Merxheim, was known as Reynobond 160 PE. It was 4mm thick and comprised a 3mm core of low-density polyethylene sandwiched between two 0.5mm aluminium sheets.

¶17.3 In January 1997, Reynobond 160 PE was tested by Warrington Fire Research Centre in accordance with BS 476 Parts 6 and 7 and achieved results which enable it to be rated Class 0.¹⁰¹⁰ {ARC00000355}; {ARC00000356}; {ARC00000357}.

¶17.4 In about 2000, Arconic began producing sheets of ACM material with a polyethylene core at its Merxheim factory.¹¹¹¹ Schmidt {Day91/13:7-15}. It was sold by Arconic under the name Reynobond 55 PE.¹²¹² Schmidt {Day91/38:17}-{Day91/39:6}. "Reynobond 55 PE" was used when the aluminium sheets were 0.5mm. Other variants, such as Reynobond 33 PE, refer to a signage product with 0.3mm aluminium sheets. The reasons for changing the name of the product are not clear. There is evidence that Reynobond 55 PE was not in all respects the same as Reynobond 160 PE, or at least that the two were not directly comparable.¹³¹³ Lane, Supplemental Phase 1 Report {BLAS0000011/33} paragraph 11.9.18; Lane, Phase 1 Report, Appendix O {BLAS0000036/103} paragraph O19.1.5. In particular, the core of the product manufactured in the United States was linear low density polyethylene (LLDP), whereas the core of the material produced by Arconic in France was low density polyethylene (LDP).¹⁴¹⁴ Declaration of Kevin Juedeman {INQ00014554} (declaration from the civil proceedings against Arconic in the Eastern District of Pennsylvania, Case No 219-cv-02664-MMB) paragraphs 6, 8, 11. Mr Schmidt's evidence initially was that he believed that the two products were the same but ultimately accepted that the chemical compositions of Reynobond 160 and Reynobond 55 PE were different.¹⁵¹⁵ Schmidt {Day91/12:22}-{Day91/13:15}; {Day93/10:3}-{Day93/11:10}; See also declaration of Kevin Juedeman {INQ00014554} (declaration from the civil proceedings against Arconic in the Eastern District of Pennsylvania, Case No 219-cv-02664-MMB) paragraphs 6, 8, 11.

¶17.5 There were differences between the coatings of Reynobond 160 PE and Reynobond 55 PE¹⁶¹⁶ Lane, Supplemental Phase 1 Report {BLAS0000011/30} paragraphs 11.9.1-11.9.18; Lane, Phase 1 Report, Appendix O {BLAS0000036/103} paragraph O19.1.5. and as a result the test data relating to Reynobond 160 PE did not apply to Reynobond 55 PE.¹⁷¹⁷ Lane, Supplemental Phase 1 Report {BLAS0000011/30} paragraphs 11.9.1-11.9.18; Lane, Phase 1 Report, Appendix O {BLAS0000036/103} paragraph O19.1.5.

¶17.6 Mr Wehrle claimed that he had no reason to suppose that there was any material difference between the fire performance of Reynobond 160 PE and that of Reynobond 55 PE,¹⁸¹⁸ Wehrle {MET00053190/14} page 14, paragraph 47. but we doubt that that can be correct. There is reason to believe that both the cores and the coatings of the two products were different and we do not think that it was reasonable for him to have assumed that the reaction to fire of the product installed on Grenfell Tower was the same as that of Reynobond 160 PE, which had been tested about 20 years earlier.

¶17.7 In any event, it was not appropriate for Arconic to rely on test data obtained in relation to a different product when assessing the performance of Reynobond 55 PE. That much is clear from BS 476-10¹⁹¹⁹ That addresses 'Fire tests on building materials and structures' and consists of a 'Guide to the principles, selection, role and application of fire testing and their outputs' {BSI00001757/20}. and from the wording of Warrington Fire's report, which expressly provided that its results were applicable only to the specimens tested.²⁰²⁰ {ARC00000355/3}; {ARC00000355/5}.

July 2006: Colin Southgate's warnings

¶17.8 On 26 July 2006, Colin Southgate, then the UK sales representative for Reynobond,²¹²¹ French {MET00053162/2} page 2, paragraph 7. sent an email to Guy Scheidecker (a senior executive), Gérard Sonntag, Claude Wehrle and others at Arconic about the position in relation to the fire certification of Reynobond.²²²² {MET00064988/19}. In it he pointed out that Reynobond 160 PE had been certified in 1997 and that the certificate was due to expire in 2007 at the latest. He also pointed out that the certificate did not apply to Reynobond 55 PE with either a PVdF or Duragloss coating. He then admitted that Arconic had provided certificates that did not apply to the products being sold to see if that would keep customers happy and expressed concern that it was misleading the market in certain respects.

¶17.9 The document shows quite clearly that by July 2006 at the latest personnel at Arconic, including senior executives, were aware that it had no valid certification of the fire performance of Reynobond 55 PE with a Duragloss 5000 finish and that its marketing materials were apt to mislead customers into thinking that some Reynobond PE products (including those with a Duragloss 5000 coating) possessed certifications that they did not have. The following day, 27 July 2006, Mr Sonntag wrote to Mr Southgate,²³²³ {MET00064988/20}. explaining that Arconic had asked Warrington Fire to carry out tests on Reynobond 33 PE and Reynobond 55 FR. No steps appear to have been taken to organise tests on Reynobond 55 PE, although Arconic was aware that the UK construction industry almost invariably purchased panels with PE cores.²⁴²⁴ Derrendinger {MET00053191/4} page 4, paragraph 17; French {Day87/67:19-24}; {Day88/195:15-16}.

¶17.10 In his statement, Mr Wehrle said that the results of the tests carried out on Reynobond 33 PE in 2006 had satisfied the requirements for Class 0.²⁵²⁵ Wehrle {MET00053190/9} page 9, paragraph 34. That gave him confidence that Reynobond 55 PE would achieve the same classification, because it had thicker aluminium skins and a proportionally lower polyethylene content.²⁶²⁶ Arconic Module 2 Closing Submissions {ARC00000770/22} paragraph 103; Overarching Closing Submissions {ARC00000797/37} paragraph 141. However, no one at Arconic explained what, if any, steps had been taken at that time to analyse the results of the tests on Reynobond 33 PE and apply them to Reynobond 55 PE. It seems unlikely that any such steps were taken but in any event the two products were different and test results on one could not be applied to the other. As Mr Schmidt accepted, the test reports relating to Reynobond 33 PE were simply irrelevant to Reynobond 55 PE.²⁷²⁷ Schmidt {Day91/44:20}-{Day91/45:1}.

¶17.11 On 30 June 2017, in the wake of the Grenfell Tower fire, Claude Wehrle sent an email to Diana Perriah,²⁸²⁸ {META00001104}. the President of Arconic Building and Construction Systems in the USA, attaching what he described as a list of all relevant certifications possessed by Arconic.²⁹²⁹ {META00001106}. The list did not contain any reference to a test carried out on Reynobond 160 PE in 1997 or to a test carried out on Reynobond 33 PE in 2006, which suggests that Mr Wehrle did not think that either was relevant to products being sold by Arconic (including Reynobond 55 PE) at that time. In our view, his assertion that historic test data relating to Reynobond 160 PE was relevant to Reynobond 55 PE was simply an attempt to justify the sale of that product.

¶17.12 Indeed, the results of the separate European tests that Arconic had carried out on Reynobond 55 PE in late 2004 made it clear how that product would react to fire. Arconic knew very well that the earlier tests on other products to which we have referred above could not be relied on to market Reynobond 55 PE. We now turn to examine those tests.

European testing: 2004 to 2006

December 2004: Tests 5A and 5B

¶17.13 In December 2004, on the instruction of Arconic, the French national testing organisation for the construction industry, the Centre Scientifique et Technique du Bâtiment ("CSTB"), conducted tests on Reynobond 55 PE in accordance with the European testing regime. The European single burning item test (EN 13823) required the product to be tested in its fabricated form, that is, in riveted or cassette form, and each test to be conducted on at least three specimens.³⁰³⁰ {BSI00000620}. See also Part 2, Chapter 5. The classification was derived from the mean of the results of the three samples. However, if the mean of those three results did not meet the required standard, two further specimens could be tested, in which case the highest and lowest results were excluded from consideration and the remaining three results used to determine the classification.³¹³¹ {BSI00000620/15} Section 7. On 2 December 2004 CSTB conducted two EN 13823 tests on Reynobond 55 PE, one on a panel which had been fabricated for riveted fixing (Test 5A), the other on a panel which had been fabricated for cassette fixing (Test 5B). It is important to note that from the outset in 2004 Reynobond PE 55 was always tested by Arconic in one of the two ways in which it was intended to be used, riveted fixing or cassette fixing, and never as a simple sheet of material in the condition in which it left the factory. That was necessary because the single burning item test is a test on a product in the form in which it is designed to be used. If Arconic had wished to test the product in the condition in which it left the factory it could have done so in accordance with BS 476-6 and 476-7, but it did not choose to do so. That is a matter of some significance in the light of the way in which Arconic described Reynobond 55 PE to the BBA and its closing statement in the Inquiry.

Test 5A

¶17.14 In Test 5A the riveted specimen achieved results that were, or appeared to be, consistent with European classification B-s2, d0.³²³² {ARC00000535}; {ARC00000358}. However, in April 2013, Mr Wehrle wrote in an internal email that the tests had not been "really reflective of the riveted system in general".³³³³ {MET00064988/121}. In June 2016, he wrote in another internal email that Arconic had achieved Euroclass B for Reynobond 55 PE in rivet form by manipulating the test (in his words, ""arranging" the system to pass").³⁴³⁴ {MET00064988/129}. We do not know what form the manipulation of the test took, but it seems clear that, by 2013 at the very latest, Mr Wehrle did not think that a B classification was truly representative of how the product in riveted form would typically behave.

¶17.15 Mr Schmidt accepted that it appeared that Mr Wehrle had known or suspected that the Euroclass B classification had not been honestly achieved.³⁵³⁵ Schmidt {Day91/76:5-9}. There is further support for the view that the test had been manipulated in the fact that the end use condition was a minimum air gap of 50mm,³⁶³⁶ Test 5A classification report dated 7 January 2005 {ARC00000358/4} paragraph 4.3. whereas in later tests it was reduced to 20mm, which was more realistic.³⁷³⁷ Wehrle {MET00053190/19} page 19, paragraph 64; Email chain between Mr Wehrle and Maxine Bauer of CSTB of 1 July 2011 {MET00053158/184}. Mr Wehrle's emails to his colleagues had become very frank by 2013 and we consider that we are justified in placing a good deal of weight on what he said. We think it more likely than not that the original Test 5A for riveted panels, which was said to justify a European Class B classification, was in some respects not representative of how the product would be used in practice. The results were therefore unreliable, as Arconic was aware.

Test 5B

¶17.16 Test 5B, in which the specimen was fixed in cassette form, was stopped after 850 seconds because the heat release rate had exceeded 400 kW.³⁸³⁸ {ARC00000536}. The test report recorded that the results could not be used but gave an idea of the way in which the product would behave in a fire.³⁹³⁹ {ARC00000536/7}. In his witness statement Mr Wehrle said that he understood that those words meant that Arconic could not use the result to get a classification because it "did not reflect the product's real fire performance"⁴⁰⁴⁰ Wehrle {MET00053190/15} page 15, paragraph 52. That strikes us as an odd way to read the report. It plainly meant that the results could not be used to obtain a B classification, but that the record of what had happened during the test provided an insight into the product's actual fire performance.

¶17.17 Neither Mr Wehrle nor anyone else at Arconic attempted to discuss the result with CSTB or carry out any further tests on the product in cassette form.⁴¹⁴¹ Wehrle {MET00053190/15} page 15, paragraph 52. Mr Wehrle said that everyone at Arconic was puzzled by the result but justified the decision not to carry out further tests by treating it as a rogue result and therefore unrepresentative. That conclusion was based mainly on the expectation that the product in cassette form would perform better than in the riveted form product because it had fewer exposed edges. Mr Wehrle said that no one else in the industry had obtained similar results, although he admitted that his understanding was based only on information in the public domain and what he had picked up in conversation.⁴²⁴² Wehrle {MET00053190/15} page 15, paragraphs 51, 53. The outcome was that Arconic's marketing, technical and management teams decided that it was permissible to claim that Reynobond 55 PE was Class B, whether in rivetted or cassette form, and no further testing in cassette form was considered at that time.

¶17.18 In our view, Arconic had no reasonable excuse for failing to confirm by further testing that the result of the only test it had carried out on the product in cassette form was indeed unrepresentative. If it had genuinely thought that to be the case, there was every reason to carry out further tests to establish that. Moreover, the expectation that cassette-fixed panels would perform better because they had fewer exposed edges ignored the fact that the aluminium skin had to be routed in order to enable them to be bent into shape, thereby increasing the total area of the exposed core. It also took no account of other factors that had the potential to affect the product's reaction to fire in cassette form, including its tendency to trap molten polyethylene and its greater propensity to deform and expose the core to the flame front because the skins are not mechanically fixed in place.

¶17.19 Mr Schmidt accepted that Reynobond 55 PE in cassette form never achieved a Class B classification.⁴³⁴³ Schmidt {Day91/3:12}-{Day91/4:3}. In any event, Arconic knew from early 2005 that the only available test evidence indicated that Reynobond 55 PE in cassette form had the potential to react to fire in an extremely dangerous way and had failed to obtain any classification under the European standards. Despite that, Arconic persisted in telling the market that Reynobond 55 PE had been classed B-s2, d0 regardless of the form in which it was used.

¶17.20 In the period leading up to the Grenfell Tower fire Arconic carried out reaction to fire tests on various forms of its Reynobond product. In none of them did the product satisfy the requirements of Class B and the results as a whole demonstrated clearly that Test 5B had been anything but "rogue". Even if, contrary to our conclusion, Arconic had believed in early 2005 that the result of Test 5B was unrepresentative, by 2011 it knew very well that it was not.

¶17.21 Arconic did not disclose the results of Test 5B to any testing organisation, certification body or end user in the UK at any time before the Grenfell Tower fire. Mr Schmidt accepted that if the fire had not taken place it was likely that they would have remained secret.⁴⁴⁴⁴ Schmidt {Day91/103:9-15}.

2003 and 2006 tests on Reynobond 55 FR (Riveted form)

¶17.22 In June 2003, Arconic arranged for Reynobond 55 with a fire-resistant core to be tested by Warrington Fire under BS 476-6 and BS 476-7. Taken together, the results of those tests were capable of supporting a national Class 0 classification for that product.⁴⁵⁴⁵ Warrington Fire Research Test Report 132317 (BS 476-6 test) {BBA00000053} and Warrington Fire Test Report 132316 (BS 476-7 test) {BBA00000050}. Reynobond with an unmodified polyethylene core was not tested at that time.

¶17.23 In October 2006, Reynobond 55 FR in rivet form with a gold-coloured coating was tested by CSTB in accordance with the EN 13823 single burning item method and achieved a classification of B-s1, d0.⁴⁶⁴⁶ CSTB – Reaction to Fire Classification Report RA06-0372 {BBA00008288}. Unsurprisingly, that was better than its PE equivalent. There is no evidence that Reynobond 55 FR was tested in cassette form at that time.